From any computer that has ClearML serving installed. It is simply used to submit configurations, not actually run anything. Please refer to the step by step setup here for more info 🙂
I'm using the helm chart, is that not part of it?
Nope! The helm chart sets up all the infrastructure to run everything. What exactly to run is decided using the clearml-serving CLI. Using it, you can swap out models, setup A/B testing of different versions, do canary rollouts etc. But the HELM stack is there only to run what you defined using the CLI
Hi! Have you run clearml-serving create ...
first? Usually you'd make what's called a "control plane task" first, that will hold all your configuration. Step 4 in the initial setup instructions is where you'll find it!